In Fourier analysis, Parseval's Identity relates to "the summability of the Fourier series as a function."
In inner product space analysis, the "identity" works as a "Pythagorean theorem" relating the squared length of a vector to the sums of squares of its components.
How would the same theorem be central to two different applications? Or are they more or less connected because the "summability" of the Fourier Series, is somehow also based on a "Pythagorean" theorem?
The Fourier theorem is in a way a special case of Parseval's identity. Consider the following: If you look at the space $L^2$ with inner product
$$\left\langle f,g \right\rangle = \frac{1}{2\pi} \int_{0}^{2\pi} f(x)\overline{g(x)} dx$$
Then the functions $e^{ikx}, k\in\mathbb{Z}$ form an orthonormal basis, so Parseval tells you $$\|f\|^2 = \sum_{k\in\mathbb{Z}} \left|\left\langle f,e^{ikx}\right\rangle\right|^2 $$
However $c_k = \left\langle f,e^{ikx}\right\rangle = \frac{1}{2\pi} \int_{0}^{2\pi} f(x)e^{-ikx} dx$ is exactly the $k$-th Fourier coefficient, so this is nothing other than Parseval's identity for Fourier series.
It all comes down to the fact, that taking the Fourier series of a function is nothing other then expressing this function in the basis given by $e^{ikx}$: $$ f = \sum_{k\in\mathbb{Z}} \left\langle f,e^{ikx}\right\rangle e^{ikx} = \sum_{k\in\mathbb{Z}} c_k e^{ikx} $$
(Your constants may vary depending on definitions)